The better we understand the structural determinants of broadly neutralizing influenza antibodies, the better vaccines we can design! Project 2 ?Innovative Computational Approaches for Structure Prediction and Design of Novel Human Influenza Antibodies? will develop a suite of in silico tools that will identify and develop such novel antibodies in conjunction with project 1 for influenza as a model system. Several high-resolution structures of antibodies engaging influenza A HA have been determined by X-ray crystallography. The availa- bility of this rich structural base and the experience of our collaborators with the system make influenza a good model system for the present proposal. Influenza is also a good model system for biodefense: H5N1 influenza viruses traditionally infect birds, but have been responsible for several recent outbreaks limited to bird-to- human transmission. Recent research described adaptations of influenza H5N1 that confer respiratory droplet transmissibility from ferret to ferret, which may mimic the future development of a highly pathogenic pandemic human H5 virus in nature. To this end we recently isolated a human neutralizing monoclonal antibody to the H5 head domain that recognizes both wild-type and respiratory droplet transmissible H5 HAs from humans vac- cinated with conventional H5 HA protein vaccine. Encouraged by this success, the first objective of this project is to develop computational algorithms that in concert with experimental approaches provided by project 1 and the structure determination core will swiftly identify and structurally characterize human antibodies that broadly neutralize influenza viruses ? an important strategy for the rapid response to emerging threats to human health. Project 1 will sequence the antibody repertoire of humans before and after being challenged with an influenza vaccine or being infected with the virus and pass these to project 2 for identification and prioritization of antibodies for functional and structural characterization in project 1 and the structural biology core. In collab- oration with the Crowe laboratory, we recently demonstrated that a new computational method termed ?multi- state design? can recapitulate antibody maturation in silico. The second objective of this project is therefore to apply the newly developed computational tools for (a) in silico maturation of antibodies to increase affinity for the HA antigen of specific virus types and (b) multi-state design to create antibodies that recognize HAs of mul- tiple different clades, subtypes, groups, or even types. Specifically, we will mature antibodies in silico with re- spect to target HA types and subtypes. Starting from one of 25 HA/antibody co-crystal structures, the HA will be replaced with an HA structure of the desired type or subtype. Then, single-state design will be employed to predict an ideal antibody sequence to bind the respective HA. Similarly, multi-state design will be applied to design novel antibodies that bind multiple HA clades, groups, subtypes and types. For both experiments, a lim- ited number of computational designs with predicted binding affinity as good as, or better than, the starting co- crystal structures will be submitted for experimental validation in project 1 and the structural biology core.